CN110298503A - Tunnel rock burst method for early warning based on microseism information and depth convolutional neural networks - Google Patents
Tunnel rock burst method for early warning based on microseism information and depth convolutional neural networks Download PDFInfo
- Publication number
- CN110298503A CN110298503A CN201910560076.2A CN201910560076A CN110298503A CN 110298503 A CN110298503 A CN 110298503A CN 201910560076 A CN201910560076 A CN 201910560076A CN 110298503 A CN110298503 A CN 110298503A
- Authority
- CN
- China
- Prior art keywords
- rock burst
- tunnel
- early
- prewarning unit
- seismic monitoring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000011435 rock Substances 0.000 title claims abstract description 291
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 39
- 238000012544 monitoring process Methods 0.000 claims abstract description 100
- 238000012360 testing method Methods 0.000 claims abstract description 18
- 238000002360 preparation method Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 18
- 238000009412 basement excavation Methods 0.000 claims description 13
- 230000000694 effects Effects 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims description 11
- 230000002411 adverse Effects 0.000 claims description 8
- 238000009395 breeding Methods 0.000 claims description 8
- 230000001488 breeding effect Effects 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 claims description 5
- 238000013519 translation Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 238000007619 statistical method Methods 0.000 claims description 4
- 238000009825 accumulation Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000011081 inoculation Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims description 3
- 239000013589 supplement Substances 0.000 claims description 3
- 230000001537 neural effect Effects 0.000 claims 1
- 230000002265 prevention Effects 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 abstract description 3
- 238000013528 artificial neural network Methods 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 239000000047 product Substances 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000005553 drilling Methods 0.000 description 2
- 239000004744 fabric Substances 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000005381 potential energy Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Operations Research (AREA)
- Primary Health Care (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Biomedical Technology (AREA)
- Entrepreneurship & Innovation (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Geophysics And Detection Of Objects (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The present invention provides the tunnel rock burst method for early warning based on microseism information and depth convolutional neural networks, is related to rock burst early warning technology field.This method is firstly, the rock burst case for establishing known rock burst grade and its micro seismic monitoring message sample library in preparation process.Then, the spatial dimension and time span of prewarning unit are selected, many reference amounts micro seismic monitoring information in prewarning unit is exported.A large amount of rock burst-microseism message sample pair is constructed by fine tuning prewarning unit.The depth convolutional neural networks that sample establishes input, pass through the best rock burst Early-warning Model of the acquisitions such as study, test, debugging.Then, by the microseism information input rock burst Early-warning Model in prewarning area prewarning unit, the grade and probability to the potential rock burst of prewarning area are obtained.The method of the present invention practicability with higher and accuracy rate provide scientific basis to formulate targetedly rock burst prevention and control measure, it can be achieved that the intelligent real-time early warning of potential rock burst and its grade.
Description
Technical field
The present invention relates to rock burst early warning technology fields, more particularly to one kind to be based on microseism information and depth convolutional neural networks
Tunnel rock burst method for early warning.
Background technique
Rock burst is under excavation or other external disturbances, and the flexible deformation potential energy built up in underground engineering rock mass is released suddenly
It puts, leads to the dynamic phenomenon of country rock explosion, ejection.As the engineerings such as mine, water conservancy and hydropower, access tunnel (road) are gradually to deep
Extend, rock burst hazard problem is more and more prominent.The generation of rock burst often brings casualties, economic loss and duration to engineering
The harm such as delay.Carrying out scientific and effective early warning to rock burst is the important channel for avoiding or reducing these harm.
The experts and scholars of recent domestic have done a lot of research work for rock burst early warning, and achieve some researchs at
Fruit.Chinese invention patent " rock mass stress develops long-term real-time monitoring and rock burst early warning visualization device and method ", publication number
CN106768510A discloses the stress variation around the rod piece monitoring drilling that a kind of utilization is installed in drilling, thus to rock
The quick-fried method for carrying out early warning, the effective monitoring scope of this method is smaller, nearby fails the region for installing the rod piece for face
It not can be carried out effectively monitoring and early warning, and face near zone is the district occurred frequently of rock burst." one kind is based on sound to Chinese invention patent
The strain type rock burst method for early warning of sound signal waveform variation characteristic ", publication number CN105676268A are disclosed a kind of by adopting
The method that voice signal in collection, record and analysis rock burst destructive process carries out strain type rock burst early warning, due to being destroyed when rock burst
Sound when having generated, rock burst will soon occur, so carrying out early warning to the implementation of rock burst prevention and control measure using this method
The reserved time is very limited.Chinese invention patent " a kind of Rock burst proneness differentiation based on bayesian theory and the pre- police
Method ", publication number CN106840843A, it is basic for disclosing one kind with rock strength index, surrouding rock stress index and energy indexes
The rock burst Early-warning Model of the factor, the model is although it is contemplated that the polynary factor, but since the physical and mechanical parameter for obtaining rock is time-consuming
Arduously, it is difficult to realize the real-time early warning of rock burst in this way.A kind of Chinese invention patent " rock burst hazard infrared thermal imagery early warning knowledge
Method for distinguishing ", publication number CN106370306A are disclosed a kind of by infrared thermal imagery equipment real-time monitoring target rock mass surface temperature
Degree, thus the method for carrying out rock burst early warning, due to construction site rock mass surface temperature divulged information, construct and ground heat affecting compared with
Greatly, early warning accuracy rate is lower in this way for institute." rock burst real-time prediction technology fills Chinese invention patent in rock tunnel work progress
Set ", publication number CN103278843A discloses a kind of rock burst method for early warning based on sound emission；A kind of Chinese invention patent " base
In the rock burst omen comprehensive and quantitative method for early warning of more microseism parameters ", publication number CN103984005A is disclosed a kind of based on microseism
The rock burst method for early warning of monitoring information.A possibility that both methods can only all occur rock burst carries out early warning, cannot be to potential
The grade of rock burst carries out early warning.It can be seen that there is also more shortcomings for existing rock burst method for early warning.
Summary of the invention
The technical problem to be solved by the present invention is in view of the above shortcomings of the prior art, provide it is a kind of based on microseism information and
The tunnel rock burst method for early warning of depth convolutional neural networks, to improve the accurate of potential rock burst early warning in tunnel engineering digging process
Property, real-time, and for formulate targetedly rock burst prevention and control measure scientific basis is provided.
In order to solve the above technical problems, the technical solution used in the present invention is: based on microseism information and depth convolution mind
Tunnel rock burst method for early warning through network, comprising the following steps:
Step 1 carries out rock burst micro seismic monitoring during tunnel excavation, obtains many reference amounts microseism in rock burst preparation process
Monitoring information；
The many reference amounts micro seismic monitoring information selects common microseism basic parameter: microseismic event number, microseism release can, it is micro-
It shakes apparent volume and the moment occurs for microseismic event, these parameters are scalars；
Step 2 carries out geologic reconnaissance and micro seismic monitoring information interpretation to the rock burst occurred, determines rock burst grade；
The rock burst grade includes no rock burst, slight rock burst, medium rock burst and strong rock burst；
Step 3, the sample database for establishing different brackets rock burst and corresponding micro seismic monitoring message sample library；
For creating underground engineering, due to lacking rock burst case and corresponding micro seismic monitoring information, geological conditions phase is selected
Like, the identical built engineering of digging mode, with the rock burst case of a large amount of known grades that occurs in its digging process and corresponding
Micro seismic monitoring information establishes sample database；After new construction is formally constructed, supervised with the rock burst case of kainogenesis and corresponding microseism
Measurement information updates, enriches database；For the engineering of hot work in progress, then with the rock burst case of accumulation early period and corresponding microseism prison
Measurement information establishes sample database；
Step 4, the sample that 80% is randomly selected in the rock burst sample database of foundation and corresponding micro seismic monitoring message sample library
This is established by the method for deep learning as test sample as training sample, the sample of residue 20% and is based on depth convolution
The tunnel rock burst Early-warning Model of neural network；
Step 4.1 determines prewarning unit；The prewarning unit is determined jointly by spatial dimension and time span；
The spatial dimension of the prewarning unit is to wrap up the cuboid of tunnel face, in central point and tunnel cross section
Heart point is overlapped；Prewarning unit follows tunnel face dynamic mobile, but remains unchanged after spatial dimension determination, the sky of prewarning unit
Between range determined by following formula:
Wherein, x is prewarning unit along the distribution in canal axes direction, and x forward direction is consistent with tunnel excavation direction, "+"
Indicate front of tunnel heading, "-" indicates face rear；Y is prewarning unit in the horizontal plane perpendicular to point in canal axes direction
Cloth range, "+" indicate that positive direction of the y-axis, "-" indicate negative direction of the y-axis；Z is distribution of the prewarning unit in vertical direction, "+"
Indicate that straight up, "-" indicates straight down；DeFor tunnel monitoring section Average equivalent diameter；
Wherein, diFor the equivalent diameter of the i-th monitoring section of tunnel；N is the hole section that tunnel monitoring section has different section size
Sum；AiFor the cross-sectional area of the i-th monitoring section of tunnel；LiFor the section girth of the i-th monitoring section of tunnel；
The time span of the prewarning unit is determining by the average inoculation duration of statistical analysis rock burst case, calculation formula
It is as follows:
Wherein, T is the time span of prewarning unit；N ' is rock burst case sum；ti′For breeding for the i-th ' a rock burst case
Duration；The eve occurred for rock burst；For the generation moment of first microseismic event in prewarning unit spatial dimension；
Generation moment and the rock burst that duration refers in prewarning unit spatial dimension first microseismic event are bred in the rock burst
The time difference of the eve of generation；Different rock burst cases breed that duration is unequal, in order to avoid the difference of duration is bred in rock burst
Adverse effect caused by learning effect to depth convolutional neural networks sets unified early warning when establishing rock burst Early-warning Model
Unit time span；
When establishing many reference amounts micro seismic monitoring information time sequence, occurred in moment or prewarning unit spatial dimension most with rock burst
The finish time that moment is prewarning unit time span occurs for the latter microseismic event；For breeding duration greater than prewarning unit
The micro seismic monitoring information of early period is bred in the rock burst case of time span, removal rock burst；When for breeding duration less than prewarning unit
Between span rock burst case, rock burst breed start before time series mend 0；The time scale of micro seismic monitoring information time sequence
As unit of hour；
Step 4.2, the fine tuning that spatial dimension and time span are carried out to determining prewarning unit；
Firstly, the spatial dimension along x-axis translation prewarning unit makes the same rock burst case correspond to multiple micro seismic monitoring information
Sample, to construct more training samples pair；Secondly, the time span of translation prewarning unit keeps the same rock burst case corresponding
Multiple micro seismic monitoring message samples, to construct more training samples pair；
Step 4.3 introduces cost matrix to eliminate or weaken sample class imbalance bring adverse effect；The rock of foundation
Cost matrix used in quick-fried Early-warning Model is as shown in table 1:
The cost matrix of 1 rock burst Early-warning Model of table
Wherein, Cost (i ", j ") is indicated j " grade rock burst is predicted as the cost of i " grade rock burst, i "=0, and 1 ..., 3,
J "=0,1 ..., 3；
Step 4.4, building depth convolutional neural networks model, and model training optimization is carried out, it obtains and is based on microseism information
With the tunnel rock burst Early-warning Model of depth convolutional neural networks；
The depth convolutional neural networks of the building include 1 input layer, 2 convolutional layers, 2 pond layers, 2 full connections
Layer and 1 Softmax layers；Input layer input many reference amounts micro seismic monitoring information time sequence, decision-making level export rock burst grade and its
Probability；
Using rock burst sample database and corresponding micro seismic monitoring message sample library to the depth convolutional neural networks model of building
It is trained optimization, according to the classification results of test sample, acquisition reaches the highest model parameter of rock burst grade separation accuracy rate,
Obtain the tunnel rock burst Early-warning Model based on microseism information and depth convolutional neural networks；
Step 5, the rock burst early warning mould that will be established to many reference amounts micro seismic monitoring information input in prewarning area prewarning unit
Type passes through the grade and its probability of the calculating output potential rock burst of prewarning area of Early-warning Model；
In the micro seismic monitoring for extracting the region using prewarning unit identical with rock burst Early-warning Model is established to prewarning area
Information establishes many reference amounts micro seismic monitoring information time sequence to prewarning area, inputs rock burst early warning as input data
The probability of each grade rock burst occurs by output after calculating to prewarning area for model, rock burst Early-warning Model；The potential rock of prewarning area
Quick-fried grade and probability is the corresponding rock burst grade of maximum probability of happening and probability in calculated result；When two or more rocks
The probability of happening of quick-fried grade differs less than 5%, and when being greater than other rock burst grade probability of happening, selects wherein highest rock burst etc.
The grade and probability of grade and its probability of happening as the potential rock burst of prewarning area；
Step 6, with the micro seismic monitoring information in the excavation of tunnel and the passage real-time update prewarning unit of time, will more
Information after new inputs rock burst Early-warning Model in real time, to carry out real-time update to early warning result；
After carrying out early warning to live rock burst, whether it is consistent by field test early warning result with actual conditions, by the secondary rock
Quick-fried and corresponding micro seismic monitoring information carries out dynamic supplement and update to original sample database as new samples, to continue to optimize rock
Quick-fried Early-warning Model.
The beneficial effects of adopting the technical scheme are that provided by the invention rolled up based on microseism information and depth
The tunnel rock burst method for early warning of product neural network, is arranged without explicit carry out characteristics extraction and threshold value, but it is implicit from
It is obtained in training sample by deep learning, can sufficiently excavate the evolution rule of micro seismic monitoring information time sequence in rock burst preparation process
Rule；Rock burst early warning accuracy rate is improved, rock burst intelligent early-warning is realized；Using many reference amounts micro seismic monitoring information time sequence conduct
Input, avoids single factors bring error, and developing rock burst early warning from the presence or absence of coarse rock burst is different brackets rock burst
Early warning；Cost matrix is introduced, the concentrated expression objective cost of rock burst early warning is gone back while solving the problems, such as sample imbalance；
Microseism information real-time update ensure that the real-time of early warning result.
Detailed description of the invention
Fig. 1 is the pre- police of the tunnel rock burst provided in an embodiment of the present invention based on microseism information and depth convolutional neural networks
Method flow chart；
Fig. 2 is tunnel rock burst prewarning unit spatial dimension schematic diagram provided in an embodiment of the present invention, wherein (a) is three-dimensional vertical
Body figure (b) is top view；
Fig. 3 is the corresponding many reference amounts micro seismic monitoring information time sequence of different brackets rock burst case provided in an embodiment of the present invention
Column, wherein (a) is no rock burst, (b) is slight rock burst, (c) is medium rock burst, (d) is strong rock burst；
Fig. 4 is that the spatial dimension provided in an embodiment of the present invention by translating prewarning unit along x-axis constructs more multisample pair
Schematic diagram, wherein (a) be prewarning unit spatial dimension 1, (b) be prewarning unit spatial dimension 2, (c) be prewarning unit space
Range 3；
Fig. 5 is the signal that the time span provided in an embodiment of the present invention by translating prewarning unit constructs more multisample pair
Figure.
In figure: 1, tunnel；2, microseismic event；3, tunnel excavation direction；4, prewarning unit spatial dimension；5, face；6,
Microseismic event number time series；7, microseism release can time series；8, microseism apparent volume time series；9, span when prewarning unit
Degree 1；10, prewarning unit time span 2；11, prewarning unit time span 3.
Specific embodiment
With reference to the accompanying drawings and examples, specific embodiments of the present invention will be described in further detail.Implement below
Example is not intended to limit the scope of the invention for illustrating the present invention.
Rock burst takes place frequently in certain tunnel engineering digging process, for reduce rock burst hazard, the present embodiment using it is of the invention based on
The tunnel rock burst method for early warning of microseism information and depth convolutional neural networks carries out the rock burst in the tunnel engineering digging process
Early warning.
Tunnel rock burst method for early warning based on microseism information and depth convolutional neural networks, as shown in Figure 1, including following step
It is rapid:
Step 1 carries out rock burst micro seismic monitoring during tunnel excavation, obtains many reference amounts microseism in rock burst preparation process
Monitoring information；
The many reference amounts micro seismic monitoring information selects common microseism basic parameter: microseismic event number, microseism release can, it is micro-
It shakes apparent volume and the moment occurs for microseismic event, these parameters are scalars；
Step 2 carries out geologic reconnaissance and micro seismic monitoring information interpretation to the rock burst occurred, determines rock burst grade；
The rock burst grade includes no rock burst, slight rock burst, medium rock burst and strong rock burst；
Step 3, the sample database for establishing different brackets rock burst and corresponding micro seismic monitoring message sample library；
It establishes abundant and accurately different brackets rock burst sample database and corresponding micro seismic monitoring message sample library is to realize
Premise of the invention.For creating underground engineering, due to lacking rock burst case and corresponding micro seismic monitoring information, geology item is selected
Part is similar, the identical built engineering of digging mode, with the rock burst case and phase of a large amount of known grades occurred in its digging process
The micro seismic monitoring information answered establishes sample database；After new construction is formally constructed, with the rock burst case of kainogenesis and corresponding micro-
Monitoring information is shaken to update, enrich database；For the engineering of hot work in progress, then with the rock burst case of accumulation early period and corresponding micro-
Shake monitoring information establishes sample database；
Step 4, the sample that 80% is randomly selected in the rock burst sample database of foundation and corresponding micro seismic monitoring message sample library
This is established by the method for deep learning as test sample as training sample, the sample of residue 20% and is based on depth convolution
The tunnel rock burst Early-warning Model of neural network；
Step 4.1 determines prewarning unit；The prewarning unit is determined jointly by spatial dimension and time span；
During establishing rock burst Early-warning Model, it is most important to choose reasonable prewarning unit；Prewarning unit is spatially
It again should include most of microseismic event of potential rock burst preparation process comprising potential rockburst risk region；Tunnel engineering excavates
It is instant type rock burst that it is highest that the frequency occurs in the process, and rock burst is frequently experienced in the section of the hole on face or near face,
And most of microseismic event also concentrates on this region, the spatial dimension in this region and the hole diameter of tunnel excavation are related；It is described
The spatial dimension of prewarning unit is to wrap up the cuboid of tunnel face, and central point is overlapped with tunnel cross-section center point, such as
Shown in Fig. 2；Prewarning unit follows tunnel face dynamic mobile, but remains unchanged after spatial dimension determination, the sky of prewarning unit
Between range determined by following formula:
Wherein, x is prewarning unit along the distribution in canal axes direction, and x forward direction is consistent with tunnel excavation direction, "+"
Indicate front of tunnel heading, "-" indicates face rear；Y is prewarning unit in the horizontal plane perpendicular to point in canal axes direction
Cloth range, "+" indicate that positive direction of the y-axis, "-" indicate negative direction of the y-axis；Z is distribution of the prewarning unit in vertical direction, "+"
Indicate that straight up, "-" indicates straight down；DeFor tunnel monitoring section Average equivalent diameter；
Wherein, diFor the equivalent diameter of the i-th monitoring section of tunnel；N is the hole section that tunnel monitoring section has different section size
Sum；AiFor the cross-sectional area of the i-th monitoring section of tunnel；LiFor the section girth of the i-th monitoring section of tunnel；
The time span of the prewarning unit is determining by the average inoculation duration of statistical analysis rock burst case, calculation formula
It is as follows:
Wherein, T is the time span of prewarning unit；N ' is rock burst case sum；ti′For breeding for the i-th ' a rock burst case
Duration；The eve occurred for rock burst；For the generation moment of first microseismic event in prewarning unit spatial dimension；
Generation moment and the rock burst that duration refers in prewarning unit spatial dimension first microseismic event are bred in the rock burst
The time difference of the eve of generation；Different rock burst cases breed that duration is unequal, in order to avoid the difference of duration is bred in rock burst
Adverse effect caused by learning effect to depth convolutional neural networks sets unified early warning when establishing rock burst Early-warning Model
Unit time span.
When establishing many reference amounts micro seismic monitoring information time sequence, occurred in moment or prewarning unit spatial dimension most with rock burst
The finish time that moment is prewarning unit time span occurs for the latter microseismic event；For breeding duration greater than prewarning unit
The micro seismic monitoring information of early period is bred in the rock burst case of time span, removal rock burst；When for breeding duration less than prewarning unit
Between span rock burst case, rock burst breed start before time series mend 0；The time scale of micro seismic monitoring information time sequence
As unit of hour, the corresponding many reference amounts micro seismic monitoring information time sequence of different grades of rock burst case is as shown in Figure 3；
Step 4.2, the fine tuning that spatial dimension and time span are carried out to determining prewarning unit；
In order to increase the quantity for the sample pair that rock burst case and its corresponding micro seismic monitoring information form, so that depth be made to roll up
Product neural network preferably learns and characterizes the universal law of each grade rock burst preparation process, can be in the sky that prewarning unit has been determined
Between prewarning unit is finely adjusted after scope and time span.
Firstly, making the same rock burst case correspond to multiple micro seismic monitorings by the spatial dimension for translating prewarning unit along x-axis
Message sample, to construct more training samples pair, as shown in Figure 4.Secondly, being made by the time span for translating prewarning unit
The same rock burst case corresponds to multiple micro seismic monitoring message samples, to construct more training samples pair, as shown in Figure 5.
Step 4.3, for same tunnel engineering, different grades of rock burst case number is different, this will cause sample class
Imbalance, to cause adverse effect to the learning effect of depth convolutional neural networks.Meanwhile rock burst early warning is that a cost is quick
Perceptual problem, economic loss when sample class imbalance will make the early warning result deviation occur tend to bigizationner.For this purpose, introducing
Cost matrix adversely affects to eliminate or weaken sample class imbalance bring；Cost used in the rock burst Early-warning Model of foundation
Matrix is as shown in table 1:
The cost matrix of 1 rock burst Early-warning Model of table
Wherein, Cost (i ", j ") is indicated j " grade rock burst is predicted as the cost of i " grade rock burst, i "=0, and 1 ..., 3,
J "=0,1 ..., 3；
Step 4.4, building depth convolutional neural networks model, and model training optimization is carried out, it obtains and is based on microseism information
With the tunnel rock burst Early-warning Model of depth convolutional neural networks；
The depth convolutional neural networks of the building include 1 input layer, 2 convolutional layers, 2 pond layers, 2 full connections
Layer and 1 Softmax layers；Input layer input many reference amounts micro seismic monitoring information time sequence, decision-making level export rock burst grade and its
Probability；
Using rock burst sample database and corresponding micro seismic monitoring message sample library to the depth convolutional neural networks model of building
It is trained optimization, according to the classification results of test sample, acquisition reaches the highest model parameter of rock burst grade separation accuracy rate,
Obtain the tunnel rock burst Early-warning Model based on microseism information and depth convolutional neural networks；
Step 5, the rock burst early warning mould that will be established to many reference amounts micro seismic monitoring information input in prewarning area prewarning unit
Type passes through the grade and its probability of the calculating output potential rock burst of prewarning area of Early-warning Model；
In the micro seismic monitoring for extracting the region using prewarning unit identical with rock burst Early-warning Model is established to prewarning area
Information establishes many reference amounts micro seismic monitoring information time sequence to prewarning area, inputs rock burst early warning as input data
The probability of each grade rock burst occurs by output after calculating to prewarning area for model, rock burst Early-warning Model；The potential rock of prewarning area
Quick-fried grade and probability is the corresponding rock burst grade of maximum probability of happening and probability in calculated result；When two or more rocks
The probability of happening of quick-fried grade differs less than 5%, and when being greater than other rock burst grade probability of happening, selects wherein highest rock burst etc.
The grade and probability of grade and its probability of happening as the potential rock burst of prewarning area；
Step 6, with the micro seismic monitoring information in the excavation of tunnel and the passage real-time update prewarning unit of time, will more
Information after new inputs rock burst Early-warning Model in real time, to carry out real-time update to early warning result；
After carrying out early warning to live rock burst, whether it is consistent by field test early warning result with actual conditions, by the secondary rock
Quick-fried and corresponding micro seismic monitoring information carries out dynamic supplement and update to original sample database as new samples, to continue to optimize rock
Quick-fried Early-warning Model.
The present embodiment with 30 in tunnel engineering digging process early period without rock burst case, 20 slight rock burst cases,
15 medium rock burst cases, 3 strong rock burst cases (totally 68 cases) and corresponding micro seismic monitoring information establish sample database.With
Machine selection 24 therein without rock burst case, 16 slight rock burst cases, 12 medium rock burst cases, 2 strong rock burst cases
(totally 54 cases) and corresponding micro seismic monitoring information form training sample pair, remaining rock burst case and corresponding micro seismic monitoring
Information forms test sample pair.
The tunnel engineering excavated section is gateway opening type, and cross dimension is 7.2m × 6.2m (wide × high), cross section face
Product is about 39.1m2, section perimeter is about 23.7m.It is about by the equivalent diameter that formula (2) and (3) can calculate the tunnel
6.6m is about [- 25,10] x ∈ by the spatial dimension that formula (1) can calculate prewarning unit, y ∈ [- 33,33], z ∈ [- 33,
33].That is prewarning unit spatially wraps up front of tunnel heading 10m, face rear 25m, in canal axes two sides 33m and tunnel
The solid space range of heart line each 33m up and down.To 68 rock burst cases to breed duration for statistical analysis, by formula (4) and
(5) time span that can calculate prewarning unit is about 144h.
For 54 training cases, first by by the spatial dimension of prewarning unit along x-axis respectively translation -5m, 0m and+
5m makes the corresponding 3 micro seismic monitoring message samples of the same rock burst case.Then again by by the time span of prewarning unit to pregnant
It educates and translates 1h, 2h and 3h respectively in early days, make the corresponding 3 micro seismic monitoring message samples of the same rock burst case.Two methods are comprehensive,
It can make the corresponding 9 micro seismic monitoring message samples of the same rock burst case, so that 486 training samples pair can be constructed altogether.
By training sample to input depth convolutional neural networks, it is allowed to more ginsengs in different grades of rock burst preparation process
It measures micro seismic monitoring information time sequence and carries out deep learning.Rock burst Early-warning Model is obtained by study, then utilizes the model pair
Test sample is to testing.According to the model parameter of test result appropriate adjustment depth convolutional neural networks, until obtaining pre-
The alert highest rock burst Early-warning Model of accuracy rate.In order to eliminate sample imbalance bring adverse effect, in rock burst Early-warning Model
Cost matrix as shown in Table 1 is introduced in trained and test process.The network structure of the rock burst Early-warning Model finally obtained
It successively include: that 1) input layer, input microseismic event number time series, microseism discharge energy time series and microseism apparent volume time sequence
Column；2) first convolutional layer, includes 16 3 × 3 convolution kernels, and convolution kernel moving step length is automatic true by same-padding strategy
It is fixed, using ReLU as activation primitive；3) first pond layer, using maximum pondization strategy；4) second convolutional layer includes 8
A 3 × 3 convolution kernel, convolution kernel moving step length is automatically determined by same-padding strategy, using ReLU as activation primitive；
5) second pond layer, using maximum pondization strategy；6) first full articulamentum includes 64 neuron nodes, and using general
The Dropout strategy that rate is 0.1, using ReLU as activation primitive；7) second full articulamentum includes 16 neuron sections
Point；8) Softmax layers, include 4 nodes, corresponding with rock burst number of degrees, and final rock burst early warning is exported using Softmax function
As a result.
Test sample is tested using above-mentioned rock burst Early-warning Model, the results are shown in Table 2.Model as shown in Table 2
Test accuracy rate has reached 85.7%.Rock burst early warning is carried out using the model in the subsequent digging process of the tunnel, is carried out in total
109 rock burst early warning, the comparison of early warning result and on-site actual situations are as shown in table 3.The present invention is to potential rock as shown in Table 3
The accuracy rate of quick-fried early warning has reached 84%.This application shows proposed by the present invention based on microseism information and depth convolutional neural networks
Tunnel rock burst method for early warning practicability with higher and accuracy rate, it can be achieved that the intelligence of potential rock burst and its grade is pre- in real time
It is alert, and scientific basis is provided to formulate targetedly rock burst prevention and control measure.
2 rock burst Early-warning Model test result of table
Table 3 is compared based on the potential quick-fried early warning result of certain tunnel engineering and on-site actual situations of the invention
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify to technical solution documented by previous embodiment, or some or all of the technical features are equal
Replacement；And these are modified or replaceed, model defined by the claims in the present invention that it does not separate the essence of the corresponding technical solution
It encloses.
Claims (8)
1. a kind of tunnel rock burst method for early warning based on microseism information and depth convolutional neural networks, it is characterised in that: including with
Lower step:
Step 1 carries out rock burst micro seismic monitoring during tunnel excavation, obtains many reference amounts micro seismic monitoring in rock burst preparation process
Information；
Step 2 carries out geologic reconnaissance and micro seismic monitoring information interpretation to the rock burst occurred, determines rock burst grade；
The rock burst grade includes no rock burst, slight rock burst, medium rock burst and strong rock burst；
Step 3, the sample database for establishing different brackets rock burst and corresponding micro seismic monitoring message sample library；
Step 4, the sample work that 80% is randomly selected in the rock burst sample database of foundation and corresponding micro seismic monitoring message sample library
For training sample, the sample of residue 20% is established by the method for deep learning as test sample and is based on depth convolutional Neural
The tunnel rock burst Early-warning Model of network；
Step 4.1 determines prewarning unit；The prewarning unit is determined jointly by spatial dimension and time span；
Step 4.2, the fine tuning that spatial dimension and time span are carried out to determining prewarning unit, keep the same rock burst case corresponding
Multiple micro seismic monitoring message samples；
Step 4.3 introduces cost matrix to eliminate or weaken sample class imbalance bring adverse effect；
Step 4.4, building depth convolutional neural networks model, and model training optimization is carried out, it obtains and is based on microseism information and depth
Spend the tunnel rock burst Early-warning Model of convolutional neural networks；
Step 5, the rock burst Early-warning Model that will be established to many reference amounts micro seismic monitoring information input in prewarning area prewarning unit, lead to
Cross the grade and its probability of the calculating output potential rock burst of prewarning area of Early-warning Model；
Step 6, with the micro seismic monitoring information in the excavation of tunnel and the passage real-time update prewarning unit of time, after update
Information input rock burst Early-warning Model in real time, thus to early warning result carry out real-time update；
After carrying out early warning to live rock burst, whether be consistent by field test early warning result with actual conditions, by the secondary rock burst and
Corresponding micro seismic monitoring information carries out dynamic supplement and update to original sample database as new samples, so that it is pre- to continue to optimize rock burst
Alert model.
2. the tunnel rock burst method for early warning according to claim 1 based on microseism information and depth convolutional neural networks,
Be characterized in that: many reference amounts micro seismic monitoring information selects common microseism basic parameter: microseismic event number, microseism release can,
Moment occurs for microseism apparent volume and microseismic event, these parameters are scalars.
3. the tunnel rock burst method for early warning according to claim 1 based on microseism information and depth convolutional neural networks,
It is characterized in that: the step 3 method particularly includes:
For creating underground engineering, due to lacking rock burst case and corresponding micro seismic monitoring information, selection geological conditions is similar, opens
The identical built engineering of excavation formula, with the rock burst case and corresponding microseism prison of a large amount of known grades occurred in its digging process
Measurement information establishes sample database；After new construction is formally constructed, with the rock burst case of kainogenesis and corresponding micro seismic monitoring information
It updates, enrich database；For the engineering of hot work in progress, then with the rock burst case and corresponding micro seismic monitoring information of accumulation early period
Establish sample database.
4. the tunnel rock burst method for early warning according to claim 1 based on microseism information and depth convolutional neural networks,
Be characterized in that: the spatial dimension of prewarning unit described in step 4.1 is to wrap up the cuboid of tunnel face, central point and tunnel
Cross-section center point is overlapped；Prewarning unit follows tunnel face dynamic mobile, but remains unchanged after spatial dimension determination, early warning
The spatial dimension of unit is determined by following formula:
Wherein, x is prewarning unit along the distribution in canal axes direction, and x forward direction is consistent with tunnel excavation direction, and "+" indicates
Front of tunnel heading, "-" indicate face rear；Y is prewarning unit in the horizontal plane perpendicular to the distribution model in canal axes direction
It encloses, "+" indicates that positive direction of the y-axis, "-" indicate negative direction of the y-axis；Z is distribution of the prewarning unit in vertical direction, and "+" indicates
Straight up, "-" indicates straight down；DeFor tunnel monitoring section Average equivalent diameter；
Wherein, diFor the equivalent diameter of the i-th monitoring section of tunnel；N is the hole section sum that tunnel monitoring section has different section size；
AiFor the cross-sectional area of the i-th monitoring section of tunnel；LiFor the section girth of the i-th monitoring section of tunnel；
The time span of the prewarning unit determines that calculation formula is such as by the average inoculation duration of statistical analysis rock burst case
Under:
Wherein, T is the time span of prewarning unit；N ' is rock burst case sum；ti′Duration is bred for the i-th ' a rock burst case；The eve occurred for rock burst；For the generation moment of first microseismic event in prewarning unit spatial dimension；
The rock burst breeds duration and refers to that the generation moment of first microseismic event and rock burst occur in prewarning unit spatial dimension
Eve time difference；Different rock burst cases breed that duration is unequal, in order to avoid rock burst breeds the difference of duration to deep
Adverse effect caused by the learning effect of convolutional neural networks is spent, unified prewarning unit is set when establishing rock burst Early-warning Model
Time span；
When establishing many reference amounts micro seismic monitoring information time sequence, last is occurred in moment or prewarning unit spatial dimension with rock burst
The finish time that moment is prewarning unit time span occurs for a microseismic event；For breeding duration greater than the prewarning unit time
The micro seismic monitoring information of early period is bred in the rock burst case of span, removal rock burst；Span when for breeding duration less than prewarning unit
The rock burst case of degree, rock burst breed the time series before starting and mend 0；The time scale of micro seismic monitoring information time sequence is with small
When be unit.
5. the tunnel rock burst method for early warning according to claim 1 based on microseism information and depth convolutional neural networks,
It is characterized in that: the step 4.2 method particularly includes:
Firstly, the spatial dimension along x-axis translation prewarning unit makes the same rock burst case correspond to multiple micro seismic monitoring message samples,
To construct more training samples pair；Secondly, to correspond to the same rock burst case multiple for the time span of translation prewarning unit
Micro seismic monitoring message sample, to construct more training samples pair.
6. the tunnel rock burst method for early warning according to claim 1 based on microseism information and depth convolutional neural networks,
Be characterized in that: cost matrix used in the rock burst Early-warning Model of foundation described in step 4.3 is as shown in table 1:
The cost matrix of 1 rock burst Early-warning Model of table
Wherein, Cost (i ", j ") is indicated j " grade rock burst is predicted as the cost of i " grade rock burst, i "=0,1 ..., 3, j "=
0,1 ..., 3.
7. the tunnel rock burst method for early warning according to claim 1 based on microseism information and depth convolutional neural networks,
Be characterized in that: the depth convolutional neural networks of building described in step 4.4 include 1 input layer, 2 convolutional layers, 2 pond layers, 2
A full articulamentum and 1 Softmax layers；Input layer inputs many reference amounts micro seismic monitoring information time sequence, and decision-making level exports rock burst
Grade and its probability；
The depth convolutional neural networks model of building is carried out using rock burst sample database and corresponding micro seismic monitoring message sample library
Training optimization, according to the classification results of test sample, acquisition reaches the highest model parameter of rock burst grade separation accuracy rate, that is, obtains
Obtained the tunnel rock burst Early-warning Model based on microseism information and depth convolutional neural networks.
8. the tunnel rock burst method for early warning according to claim 1 based on microseism information and depth convolutional neural networks,
It is characterized in that: the step 5 method particularly includes:
The micro seismic monitoring information in the region is being extracted using prewarning unit identical with rock burst Early-warning Model is established to prewarning area,
The many reference amounts micro seismic monitoring information time sequence to prewarning area is established, inputs rock burst Early-warning Model as input data,
The probability of each grade rock burst occurs by output after calculating to prewarning area for rock burst Early-warning Model；The potential rock burst of prewarning area etc.
Grade and probability are the corresponding rock burst grade of maximum probability of happening and probability in calculated result；When two or more rock burst grades
Probability of happening differ less than 5%, and when being greater than other rock burst grade probability of happening, selection wherein highest rock burst grade and its
Grade and probability of the probability of happening as the potential rock burst of prewarning area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910560076.2A CN110298503B (en) | 2019-06-26 | 2019-06-26 | Tunnel rock burst early warning method based on microseismic information and deep convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910560076.2A CN110298503B (en) | 2019-06-26 | 2019-06-26 | Tunnel rock burst early warning method based on microseismic information and deep convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110298503A true CN110298503A (en) | 2019-10-01 |
| CN110298503B CN110298503B (en) | 2022-11-25 |
Family
ID=68028885
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910560076.2A Active CN110298503B (en) | 2019-06-26 | 2019-06-26 | Tunnel rock burst early warning method based on microseismic information and deep convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110298503B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111520193A (en) * | 2020-05-15 | 2020-08-11 | 中铁十六局集团有限公司 | Non-contact tunnel engineering construction rock burst real-time forecasting method |
| CN111520192A (en) * | 2020-05-15 | 2020-08-11 | 中铁十六局集团有限公司 | Non-contact tunnel engineering construction rock burst real-time forecasting optimization method |
| CN113156492A (en) * | 2021-04-21 | 2021-07-23 | 中国科学院武汉岩土力学研究所 | Real-time intelligent early warning method applied to TBM tunnel rockburst disasters |
| CN113221471A (en) * | 2021-06-15 | 2021-08-06 | 贵州大学 | Rock burst grade discrimination method based on improved variable fuzzy set and neural network |
| CN115263432A (en) * | 2022-09-02 | 2022-11-01 | 河海大学 | Stability and safety monitoring and analyzing system for deeply-buried tunnel |
| CN115526036A (en) * | 2022-09-19 | 2022-12-27 | 长安大学 | Method and system for judging rock burst scale grade |
| CN116973549A (en) * | 2023-09-20 | 2023-10-31 | 中铁四局集团有限公司 | Rock burst prediction method and system |
| CN117348075A (en) * | 2023-09-18 | 2024-01-05 | 中铁二院工程集团有限责任公司 | Method and device for judging low-burial rock burst of river valley stress field |
| CN117852711A (en) * | 2024-01-10 | 2024-04-09 | 昆明理工大学 | Rock burst intensity prediction method and system based on BOA-ensemble learning |
| CN117890973A (en) * | 2024-01-17 | 2024-04-16 | 核工业北京化工冶金研究院 | Rock burst tendency grade discrimination method based on occurrence frequency of earthquake events |
| CN118167430A (en) * | 2024-04-03 | 2024-06-11 | 江河安澜工程咨询有限公司 | TBM tunnel rock burst dynamic early warning system and method based on multi-source information fusion |
| US12020126B2 (en) | 2021-04-08 | 2024-06-25 | International Business Machines Corporation | Automated pressure level detection and correction |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001141835A (en) * | 1999-11-16 | 2001-05-25 | Nishimatsu Constr Co Ltd | Method for exploring bedrock before cutting face and method for confirming state of cutting face |
| WO2013134427A2 (en) * | 2012-03-06 | 2013-09-12 | Ion Geophysical Corporation | Model predicting fracturing of shale |
| CN104732070A (en) * | 2015-02-27 | 2015-06-24 | 广西大学 | Rockburst grade predicting method based on information vector machine |
| CN106054243A (en) * | 2016-06-03 | 2016-10-26 | 河南理工大学 | Rockburst multi-index prediction method based on micro-seismic monitoring |
-
2019
- 2019-06-26 CN CN201910560076.2A patent/CN110298503B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001141835A (en) * | 1999-11-16 | 2001-05-25 | Nishimatsu Constr Co Ltd | Method for exploring bedrock before cutting face and method for confirming state of cutting face |
| WO2013134427A2 (en) * | 2012-03-06 | 2013-09-12 | Ion Geophysical Corporation | Model predicting fracturing of shale |
| CN104732070A (en) * | 2015-02-27 | 2015-06-24 | 广西大学 | Rockburst grade predicting method based on information vector machine |
| CN106054243A (en) * | 2016-06-03 | 2016-10-26 | 河南理工大学 | Rockburst multi-index prediction method based on micro-seismic monitoring |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111520192A (en) * | 2020-05-15 | 2020-08-11 | 中铁十六局集团有限公司 | Non-contact tunnel engineering construction rock burst real-time forecasting optimization method |
| CN111520192B (en) * | 2020-05-15 | 2021-10-22 | 中铁十六局集团有限公司 | Non-contact tunnel engineering construction rock burst real-time forecasting optimization method |
| CN111520193A (en) * | 2020-05-15 | 2020-08-11 | 中铁十六局集团有限公司 | Non-contact tunnel engineering construction rock burst real-time forecasting method |
| US12020126B2 (en) | 2021-04-08 | 2024-06-25 | International Business Machines Corporation | Automated pressure level detection and correction |
| CN113156492A (en) * | 2021-04-21 | 2021-07-23 | 中国科学院武汉岩土力学研究所 | Real-time intelligent early warning method applied to TBM tunnel rockburst disasters |
| CN113221471A (en) * | 2021-06-15 | 2021-08-06 | 贵州大学 | Rock burst grade discrimination method based on improved variable fuzzy set and neural network |
| CN115263432A (en) * | 2022-09-02 | 2022-11-01 | 河海大学 | Stability and safety monitoring and analyzing system for deeply-buried tunnel |
| CN115263432B (en) * | 2022-09-02 | 2024-07-26 | 河海大学 | Stability and safety monitoring and analyzing system for deep-buried tunnel |
| CN115526036A (en) * | 2022-09-19 | 2022-12-27 | 长安大学 | Method and system for judging rock burst scale grade |
| CN117348075A (en) * | 2023-09-18 | 2024-01-05 | 中铁二院工程集团有限责任公司 | Method and device for judging low-burial rock burst of river valley stress field |
| CN116973549B (en) * | 2023-09-20 | 2024-01-30 | 中铁四局集团有限公司 | Rock burst prediction method and system |
| CN116973549A (en) * | 2023-09-20 | 2023-10-31 | 中铁四局集团有限公司 | Rock burst prediction method and system |
| CN117852711A (en) * | 2024-01-10 | 2024-04-09 | 昆明理工大学 | Rock burst intensity prediction method and system based on BOA-ensemble learning |
| CN117890973A (en) * | 2024-01-17 | 2024-04-16 | 核工业北京化工冶金研究院 | Rock burst tendency grade discrimination method based on occurrence frequency of earthquake events |
| CN118167430A (en) * | 2024-04-03 | 2024-06-11 | 江河安澜工程咨询有限公司 | TBM tunnel rock burst dynamic early warning system and method based on multi-source information fusion |
| CN118167430B (en) * | 2024-04-03 | 2024-10-15 | 江河安澜工程咨询有限公司 | TBM tunnel rock burst dynamic early warning system and method based on multi-source information fusion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110298503B (en) | 2022-11-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110298503A (en) | Tunnel rock burst method for early warning based on microseism information and depth convolutional neural networks | |
| Xue et al. | Predicting rockburst with database using particle swarm optimization and extreme learning machine | |
| CN109611087B (en) | Volcanic oil reservoir parameter intelligent prediction method and system | |
| CN104732070B (en) | A kind of rock burst grade prediction technique based on information vector machine | |
| US20210026031A1 (en) | Dolomite reservoir prediction method and system based on well and seismic combination, and storage medium | |
| CN108985340A (en) | Shield construction ground predicting subsidence method based on dual model fusion | |
| CN113821977B (en) | Rock burst risk assessment system and method for TBM tunnel construction | |
| CN109241627A (en) | The dynamic shoring method of probability hierarchical and the device of Automated Design supporting scheme | |
| CN112232522B (en) | Intelligent recommendation and dynamic optimization method for deep roadway support scheme | |
| CN109100790A (en) | Artificial crack simulation method and device | |
| CN114036841A (en) | Landslide incidence prediction method and system based on semi-supervised support vector machine model | |
| CN112711904A (en) | Blasting vibration characteristic parameter prediction method based on SA-GA-BP | |
| CN104390534B (en) | A kind of tunnel photoface exploision method of quality control | |
| CN115017791A (en) | Tunnel surrounding rock grade identification method and device | |
| CN116595335A (en) | Rock burst microseismic parameter multi-element multi-time-step output prediction model | |
| CN116992694B (en) | Tunnel blasting quality evaluation and optimization method | |
| CN113156492B (en) | Real-time intelligent early warning method applied to TBM tunnel rockburst disasters | |
| CN117633512A (en) | Reservoir porosity prediction method based on one-dimensional convolution gating circulation network | |
| CN117033940A (en) | Surface mine blasting vibration speed prediction method based on Pearson-CNN model | |
| Zeng et al. | Local coupled Markov chain model for simulating varied stratigraphy | |
| CN106599511B (en) | Method for optimizing coal mining method of thin coal seam longwall fully mechanized coal mining face | |
| Jiang et al. | Prediction of BlastInduced Ground Vibration (BIGV) of Metro Construction Using Difference Evolution AlgorithmOptimized Gaussian Process (DE‐GP) | |
| CN114218852B (en) | Neural network depth prediction method based on disturbance correction | |
| Li et al. | Influence Analysis of Rock Mass Mechanical Properties on Tunnel Rock Stability Based on Sensor Data Integration | |
| CN116305501B (en) | Method and system for judging influence subarea of open cut tunnel near shield tunnel collaborative construction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |